Sean Yo +ai Application · Senior Technical Program Manager, AI/ML Google Workspace

I turn emerging technology into systems people can trust.

I lead complex technical programs from ambiguous opportunity through system definition, cross-functional development, iterative validation, and demonstrable outcomes—across enterprise platforms, applied AI, and human-centred products.

Selected work ↓ Résumé Cover letter Career profile Get in touch

15Mmonthly active users in the D2L product ecosystem

$10M+product line responsibility

30+engineers across four distributed teams

0→1platforms, research programs, and products

Selected work

Three ways of making the new real.

Each program began with uncertainty. Each required a different combination of product definition, technical judgment, organizational alignment, and evidence.

01 · Model-backed learning

Syllentra AI

How could an LMS use course context, specialized agents, and human review to improve learning content safely?

Syllentra is a human-governed, multi-agent learning-platform capstone. Its five-layer product progression moved from an LMS foundation through content-aware AI, role-specific experiences, and a multi-agent quality process.

GenAI agentsLocal inferenceHuman oversightCapstone

My role

Industry advisor and capstone cohort leader: shaping the product model, use cases, AI workflows, validation scenarios, model/API decisions, and intended outcomes while preserving the student team's engineering ownership.

The system

React and TypeScript, NestJS, PostgreSQL, Prisma, Docker, and local Ollama inference. Specialized agents review structure, clarity, learning alignment, and policy, then feed a synthesis stage.

Demonstrated maturity

Architecture completed and advanced into implementation and model integration, with local inference used for development and testing. Presented as capstone-scale work—not a production model launch.

02 · Applied research

Waterloo Pedestrian Analytics

How could movement data become explorable evidence for campus planning rather than a static count?

A progressive research program combined sensor-based data collection, spatial modelling, digital-twin visualization, and augmented-reality exploration to make patterns of movement tangible.

SensorsDigital twinUnityImmersive analytics

My role

As Founding Director of CVRI, I created the applied-AI research stream and brought together the research vision, multidisciplinary teams, institutional context, and progressive validation approach.

The program

The team began with infrared ingress-and-egress counting, exposed limitations through testing, modelled the room in Unity, and developed a proof-of-concept for visualizing plausible movement patterns.

Demonstrated outcome

Conestoga's public record reports a successful proof-of-concept platform visualizing foot-traffic data in an augmented-reality environment for planning, accessibility, and resource-allocation use cases.

Read the public project record →

03 · Product evolution

Celebrating Wins

How might everyday accomplishments become useful evidence—connected to objectives, reflection, and organizational learning?

An industry-sponsored capstone evolved from Ask Josh, an assistant centred on one facilitator, into a broader system for capturing, organizing, evidencing, and celebrating accomplishments.

Product discoveryIndustry partnershipEvidence systemsWorking prototype

My role

Helped frame the proposal and product model, established the sponsor relationship, supported team check-ins and testing, and guided the shift from a person-specific assistant toward a general evidence product.

The product

Structured win capture, connections to objectives and supporting evidence, personal dashboards, reporting, PDF export, and voice-oriented capture affordances.

Demonstrated maturity

A functional public prototype demonstrated the product direction. It is presented here as industry-facing product leadership, not as a production-ready AI system.

Open the prototype →

Platform scale

I know how to land complicated things.

At D2L, I led complex cross-functional programs inside a learning ecosystem serving approximately 15 million monthly active users. The work spanned product, engineering, architecture, operations, APIs, interoperability, analytics, and enterprise adoption.

I led the 0→1 development and launch of D2L's first big-data and event platform, coordinated more than 30 engineers across four distributed teams, and helped establish D2L's first production system on AWS.

0→1 data architecture → distributed engineering → cloud production → API/platform surface → launch → scaled product organization

Operating approach

Technical ambition with retained human authority.

AI systems become trustworthy through choices made across architecture, evaluation, governance, and the everyday experience of the people who use them.

Define the system.

Make the actors, boundaries, dependencies, assumptions, and intended outcomes legible before complexity hardens around them.

Create evidence.

Move through prototypes and validation deliberately. Treat uncertainty as something to test, not something to disguise with confident language.

Keep people authoritative.

Design AI to extend judgment and capability while retaining clear ownership, review, accountability, and the right to say no.

Application documents

The two documents.

Two tailored documents connecting the role's AI requirements to demonstrated technical-program leadership, platform scale, and current GenAI work.

Résumé

Two pages focused on complex program portfolios, GenAI agents and AI features, distributed engineering, platform launches, and cross-functional leadership at scale.

Open the résumé →

Cover letter

Why Workspace, what I would bring, and a precise account of the model-launch qualification without overstating the evidence.

Open the cover letter →

Let's talk

Building AI people can use—and organizations can stand behind.

sean.yo@gmail.com · LinkedIn · seanyo.ca